4 Sep 2018 Oak Ridge - Computer scientists at Oak Ridge National Laboratory (ORNL) have developed an open source software platform that allows quantum programmes to run on multiple quantum computers regardless of their unique architecture.
Quantum computers promise to revolutionize modern computing. Their systems are based on quantum physics, where each quantum bit, or qubit, represents both a 1 and a 0 at the same time.
"Several companies have developed quantum computers, and each has associated software that users must learn in order to execute programmes", stated Alex McCaskey of ORNL's Quantum Computing Institute. "To enable application portability, we have created a heterogeneous programming model called XACC that allows quantum acceleration within standard or high-performance computing software workflows in a quantum language and hardware independent manner."
ORNL scientists leveraged XACC as part of the first successful simulation of an atomic nucleus using a quantum computer. Details of the XACC programming model were published in SoftwareX .
While this development is currently making waves across the quantum computing and nuclear communities, the rise of computing as a third pillar next to experiment and theory in low-energy nuclear physics can be traced back to 2006. It was then that an ORNL team led by David Dean, now ORNL's Associate Laboratory Director for Physical Sciences, participated in the SciDAC-2 Universal Nuclear Energy Density Functional (UNEDF) project that aimed to develop a universal density functional for the description of atomic nuclei. UNEDF was jointly funded by DOE's Advanced Scientific Computing Research (ASCR) and Nuclear Physics (NP) programmes via the Scientific Discovery through Advanced Computing (SciDAC) programme.
Since then, ORNL's Nuclear Theory Group has been a key participant in the SciDAC-3 NUclear Computational Low Energy Initiative (NUCLEI) and SciDAC-4 NUCLEI-2 projects. These projects aimed to advance computational nuclear structure, reactions, and neutrino-nucleus interactions, spurring knowledge across physics and computing and further developing the unique symbiosis between the two disciplines at ORNL. The knowledge gained from this longstanding nuclear physics research enabled the computational approaches necessary to formulate the quantum algorithm used in the successful simulation.
However, the final version of the algorithm would not have been possible without the support of ASCR's Quantum Algorithm Teams (QAT) and Quantum Testbed Pathfinder (QTP) programmes. QAT addresses the development and optimization of quantum simulation and quantum machine learning algorithms of interest to DOE's Office of Science; QTP is an effort to verify and validate scientific applications on different quantum hardware types.
The quantum simulation featured a cross-disciplinary team with representatives from ORNL's Quantum Information Science Group, the Scientific Computing Group at the Oak Ridge Leadership Computing Facility, and ORNL's Nuclear Theory Group, with support from the QAT, QTP, and SciDAC-4 NUCLEI-2.
Specifically, the team used Cloud-based IBM QX5 and Rigetti 19Q quantum computers, freely available PyQuil software, and a library designed for producing programmes in the quantum instruction language to perform more than 700,000 quantum computing "measurements" of the energy of a deuteron, the nuclear bound state of a proton and a neutron - and the simplest composite atomic nucleus. It was from these measurements that the team was able to calculate the deuteron's binding energy, or the minimum amount of energy needed to disassemble it into these subatomic particles.
The calculation was relatively simple from a physics perspective, making it an ideal task for quantum computing, a technology still in its relative infancy. But the nuclear physics knowledge necessary to port the calculation to such a novel architecture required years of collaboration and created the collaborative ORNL ecosystem needed to achieve such a breakthrough.
The continuous support by SciDAC helped build the foundations for an ecosystem that made progress such as the quantum simulation possible. "The software, theoretical and algorithmic developments, supercomputers, these all benefitted in some way from those SciDAC programmes, though we may be using them differently than we once expected", stated David Dean.